fix: agent install.sh 404 + 三阶段构建编译agent二进制

问题: 构建上下文 ./backend 无法访问 ../agent, 导致容器内
/app/agent/deploy/install.sh 不存在, 返回 404

修复:
- Dockerfile 改为三阶段构建 (backend go1.21 + agent go1.24 交叉编译 + alpine runtime)
- docker-compose backend build context 改为项目根
- routes.go /agent/arch 安装命令改为动态获取 URL
- 新增 .dockerignore
- agent_install.go: SSH 远程安装 handler
- AgentWizard.tsx: 复制按钮 fallback + 动态 server URL
- ServerTableRow.tsx: agent 状态判断

验证: curl|bash 安装成功, agent 上报数据到 MySQL ✅

.dockerignore

@@ -0,0 +1,29 @@
+# Git
+.git
+.gitignore
+
+# 前端构建产物和依赖（后端构建不需要）
+frontend/node_modules
+frontend/dist
+frontend/.cache
+
+# 日志
+**/*.log
+backend/logs/
+
+# 敏感配置
+backend/config.yaml
+**/.env
+
+# Docker 自身
+docker-compose.yml
+Dockerfile
+**/Dockerfile
+
+# 文档
+*.md
+docs/
+
+# IDE
+.idea
+.vscode

backend/Dockerfile

@@ -1,60 +1,82 @@
-# 构建阶段
+# ============================================================
+# AIOps Backend + Agent 多阶段 Dockerfile
+# Stage 1: 编译后端 (Go 1.21)
+# Stage 2: 交叉编译 Agent (Go 1.24, linux-amd64/arm64 + darwin-arm64)
+# Stage 3: 运行时 (Alpine)
+# 署名 Logan.Li
+# ============================================================
+
+# ==================== Stage 1: Backend Builder ====================
 FROM golang:1.21-alpine AS builder
 
-# 设置工作目录
 WORKDIR /app
 
-# 安装构建依赖
 RUN apk add --no-cache git ca-certificates tzdata
 
-# 设置时区
 ENV TZ=Asia/Shanghai
-
-# 设置 Go 模块代理（国内加速）
 ENV GOPROXY=https://goproxy.cn,direct
 ENV GOSUMDB=off
 
-# 复制 go.mod 和 go.sum
+# 复制后端 go.mod（利用 Docker 层缓存）
-COPY go.mod go.sum ./
+COPY backend/go.mod backend/go.sum ./
-
-# 下载依赖
 RUN go mod download
 
-# 复制源代码
+# 复制后端源码
-COPY . .
+COPY backend/ ./
 
-# 构建应用
+# 构建后端二进制
 RUN CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build \
-    -ldflags="-w -s \
+    -ldflags="-w -s" \
-    -X main.Version=$(git describe --tags --always --dirty 2>/dev/null || echo 'dev') \
-    -X main.BuildTime=$(date +%Y-%m-%d_%H:%M:%S) \
-    -X main.GitCommit=$(git rev-parse --short HEAD 2>/dev/null || echo 'unknown')" \
     -o ezops ./cmd/
 
-# 运行阶段
+# ==================== Stage 2: Agent Builder ====================
+FROM golang:1.24-alpine AS agent-builder
+
+WORKDIR /agent
+
+ENV GOPROXY=https://goproxy.cn,direct
+ENV GOSUMDB=off
+
+# 复制 agent go.mod
+COPY agent/go.mod agent/go.sum ./
+RUN go mod download
+
+# 复制 agent 源码
+COPY agent/ ./
+
+# 交叉编译三个平台的 agent 二进制
+RUN CGO_ENABLED=0 GOOS=linux  GOARCH=amd64 go build -ldflags="-s -w" -o bin/aiops-agent-linux-amd64  . && \
+    CGO_ENABLED=0 GOOS=linux  GOARCH=arm64 go build -ldflags="-s -w" -o bin/aiops-agent-linux-arm64  . && \
+    CGO_ENABLED=0 GOOS=darwin GOARCH=arm64 go build -ldflags="-s -w" -o bin/aiops-agent-darwin-arm64 .
+
+# ==================== Stage 3: Runtime ====================
 FROM alpine:latest
 
-# 安装运行时依赖
+# 安装运行时依赖（curl=健康检查，openssh-client=SSH远程安装agent）
-RUN apk --no-cache add ca-certificates tzdata curl
+RUN apk --no-cache add ca-certificates tzdata curl openssh-client
 
-# 设置时区
 ENV TZ=Asia/Shanghai
 
-# 设置工作目录
 WORKDIR /app
 
-# 复制构建产物
+# 复制后端二进制
 COPY --from=builder /app/ezops .
 
-# 创建日志目录
+# 创建运行时目录
-RUN mkdir -p /app/logs /app/uploads
+RUN mkdir -p /app/logs /app/uploads /app/agent/bin /app/agent/deploy
+
+# 复制 agent 二进制（三个平台）
+COPY --from=agent-builder /agent/bin/aiops-agent-linux-amd64  /app/agent/bin/
+COPY --from=agent-builder /agent/bin/aiops-agent-linux-arm64  /app/agent/bin/
+COPY --from=agent-builder /agent/bin/aiops-agent-darwin-arm64 /app/agent/bin/
+
+# 复制 agent 安装脚本和 systemd service 模板
+COPY agent/deploy/install.sh          /app/agent/deploy/
+COPY agent/deploy/aiops-agent.service /app/agent/deploy/
 
-# 暴露端口
 EXPOSE 8080
 
-# 健康检查
+HEALTHCHECK --interval=30s --timeout=3s --start-period=10s --retries=3 \
-HEALTHCHECK --interval=30s --timeout=3s --start-period=5s --retries=3 \
     CMD curl -f http://localhost:8080/health || exit 1
 
-# 启动应用
 CMD ["./ezops"]

backend/internal/api/handlers/agent_install.go

@@ -0,0 +1,146 @@
+package handlers
+
+import (
+	"fmt"
+	"log"
+	"net/http"
+	"strconv"
+	"time"
+
+	"ezops/internal/models"
+
+	"github.com/gin-gonic/gin"
+	"golang.org/x/crypto/ssh"
+	"gorm.io/gorm"
+)
+
+// AgentInstallHandler Agent 远程安装处理器 // Logan.Li
+type AgentInstallHandler struct {
+	db *gorm.DB
+}
+
+func NewAgentInstallHandler(db *gorm.DB) *AgentInstallHandler {
+	return &AgentInstallHandler{db: db}
+}
+
+// sshDial 复用 webshell.go 的 SSH 连接逻辑
+func sshDial(server models.Server) (*ssh.Client, error) {
+	config := &ssh.ClientConfig{
+		User:            server.Username,
+		HostKeyCallback: ssh.InsecureIgnoreHostKey(),
+		Timeout:         10 * time.Second,
+	}
+
+	switch server.AuthType {
+	case "key":
+		signer, err := ssh.ParsePrivateKey([]byte(server.SSHKey))
+		if err != nil {
+			config.Auth = []ssh.AuthMethod{ssh.Password(server.Password)}
+		} else {
+			config.Auth = []ssh.AuthMethod{ssh.PublicKeys(signer)}
+		}
+	default:
+		config.Auth = []ssh.AuthMethod{ssh.Password(server.Password)}
+	}
+
+	addr := server.IP + ":" + strconv.Itoa(server.Port)
+	return ssh.Dial("tcp", addr, config)
+}
+
+// runSSHCommand 在远程主机执行命令并返回输出
+func runSSHCommand(client *ssh.Client, cmd string) (string, error) {
+	session, err := client.NewSession()
+	if err != nil {
+		return "", err
+	}
+	defer session.Close()
+
+	output, err := session.CombinedOutput(cmd)
+	return string(output), err
+}
+
+// InstallAgent 远程安装 Agent 到指定服务器
+// POST /servers/:id/install-agent
+func (h *AgentInstallHandler) InstallAgent(c *gin.Context) {
+	id, _ := strconv.Atoi(c.Param("id"))
+
+	var server models.Server
+	if err := h.db.First(&server, id).Error; err != nil {
+		c.JSON(http.StatusNotFound, models.Response{Code: 404, Message: "服务器不存在"})
+		return
+	}
+
+	if server.Status != "running" {
+		c.JSON(http.StatusBadRequest, models.Response{Code: 400, Message: "服务器未运行，无法安装 Agent"})
+		return
+	}
+
+	// 1. SSH 连接目标主机
+	client, err := sshDial(server)
+	if err != nil {
+		log.Printf("[AgentInstall] SSH连接失败 %s: %v", server.IP, err)
+		c.JSON(http.StatusOK, models.Response{
+			Code:    500,
+			Message: "SSH 连接失败: " + formatSSHError(err),
+		})
+		return
+	}
+	defer client.Close()
+
+	// 2. 检测系统架构
+	archOut, _ := runSSHCommand(client, "uname -m")
+	arch := "amd64"
+	if len(archOut) > 0 && (contains(archOut, "aarch64") || contains(archOut, "arm64")) {
+		arch = "arm64"
+	}
+
+	// 3. 构造安装命令（使用 Agent API 服务地址）
+	// 从请求头获取后端可访问地址
+	serverURL := fmt.Sprintf("http://%s", c.Request.Host)
+	token := c.Query("token")
+
+	installCmd := fmt.Sprintf(
+		`curl -fsSL %s/agent/install.sh | sudo bash -s -- --server %s --token %s`,
+		serverURL, serverURL, token,
+	)
+
+	// 4. 执行安装
+	installOut, err := runSSHCommand(client, installCmd)
+	if err != nil {
+		c.JSON(http.StatusOK, models.Response{
+			Code:    500,
+			Message: "安装命令执行失败（可能需要 sudo 权限或 curl 未安装）",
+			Data:    installOut,
+		})
+		return
+	}
+
+	// 5. 验证安装是否成功
+	verifyOut, _ := runSSHCommand(client, "systemctl is-active aiops-agent 2>/dev/null || echo 'not_found'")
+	installed := contains(verifyOut, "active") || contains(verifyOut, "running")
+
+	if installed {
+		// 更新服务器 Agent 状态
+		h.db.Model(&server).Updates(map[string]interface{}{
+			"agent_status":  "online",
+			"agent_version": "installed",
+			"last_check_at": time.Now(),
+		})
+	}
+
+	c.JSON(http.StatusOK, models.Response{
+		Code:    200,
+		Message: func() string {
+			if installed {
+				return "Agent 安装成功，已自动启动并设置开机自启"
+			}
+			return "安装命令已执行，但 Agent 服务未检测到运行（请查看日志排查）"
+		}(),
+		Data: gin.H{
+			"installed": installed,
+			"arch":      arch,
+			"output":    installOut,
+			"verify":    verifyOut,
+		},
+	})
+}

backend/internal/api/routes/routes.go

@@ -4,6 +4,7 @@ import (
 	"ezops/internal/api/handlers"
 	"ezops/internal/api/middleware"
 	"ezops/internal/config"
+	"fmt"
 
 	"github.com/gin-gonic/gin"
 	"go.uber.org/zap"
@@ -28,6 +29,19 @@ func Register(r *gin.Engine, db *gorm.DB, log *zap.Logger, cfg *config.Config) {
 	r.StaticFile("/agent/aiops-agent-linux-arm64", "./agent/bin/aiops-agent-linux-arm64")
 	r.StaticFile("/agent/aiops-agent-darwin-arm64", "./agent/bin/aiops-agent-darwin-arm64") // macOS Apple Silicon // Logan.Li
 	r.GET("/agent/arch", func(c *gin.Context) {
+		// 动态构建安装命令 URL（根据请求来源，兼容 nginx 反代）
+		scheme := "http"
+		if c.Request.TLS != nil {
+			scheme = "https"
+		}
+		if proto := c.GetHeader("X-Forwarded-Proto"); proto != "" {
+			scheme = proto
+		}
+		host := c.Request.Host
+		if host == "" {
+			host = "localhost:8080"
+		}
+		base := fmt.Sprintf("%s://%s", scheme, host)
 		c.JSON(200, gin.H{
 			"version": "1.0.0",
 			"downloads": gin.H{
@@ -35,7 +49,7 @@ func Register(r *gin.Engine, db *gorm.DB, log *zap.Logger, cfg *config.Config) {
 				"linux-arm64":  "/agent/aiops-agent-linux-arm64",
 				"darwin-arm64": "/agent/aiops-agent-darwin-arm64",
 			},
-			"install": "curl -fsSL http://<server>:8080/agent/install.sh | sudo bash -s -- --server http://<server>:8080 --token <token>",
+			"install": fmt.Sprintf("curl -fsSL %s/agent/install.sh | sudo bash -s -- --server %s --token <token>", base, base),
 		})
 	})
 
@@ -122,6 +136,10 @@ func Register(r *gin.Engine, db *gorm.DB, log *zap.Logger, cfg *config.Config) {
 		servers.DELETE("/:id", serverHandler.DeleteServer)
 		servers.POST("/batch-delete", serverHandler.BatchDeleteServers)
 		servers.POST("/batch", serverHandler.BatchCreateServers)
+
+		// Agent 远程安装 // Logan.Li
+		agentInstallHandler := handlers.NewAgentInstallHandler(db)
+		servers.POST("/:id/install-agent", agentInstallHandler.InstallAgent)
 	}
 
 	// 云账号路由

docker-compose.yml

@@ -50,8 +50,8 @@ services:
   # ======================== Go 后端服务 ========================
   backend:
     build:
-      context: ./backend
+      context: .
-      dockerfile: Dockerfile
+      dockerfile: backend/Dockerfile
     container_name: aiops-backend
     restart: unless-stopped
     ports:

frontend/nginx.conf

@@ -48,6 +48,14 @@ server {
         proxy_read_timeout 86400;
     }
 
+    # Agent 安装脚本和二进制文件代理 → 后端
+    location /agent/ {
+        proxy_pass http://backend:8080;
+        proxy_set_header Host $host;
+        proxy_set_header X-Real-IP $remote_addr;
+        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+    }
+
     # 健康检查代理
     location /health {
         proxy_pass http://backend:8080;

frontend/src/components/agent/AgentWizard.tsx

@@ -17,11 +17,12 @@ interface AgentWizardProps {
   onOpenChange: (open: boolean) => void;
 }
 
-/** 自动检测服务器地址：使用当前浏览器访问的主机名，端口固定 8080 */
+/** 自动检测服务器地址：使用当前浏览器访问的地址（协议+主机+端口） */
 function detectServerAddress(): string {
   if (typeof window !== 'undefined' && window.location) {
-    const host = window.location.hostname;
+    // 直接用当前浏览器的 origin（如 http://10.10.10.150），
-    return host ? `http://${host}:8080` : 'http://127.0.0.1:8080';
+    // Nginx 会将 /agent/ 请求代理到后端
+    return window.location.origin;
   }
   return 'http://127.0.0.1:8080';
 }
@@ -35,12 +36,27 @@ export function AgentWizard({ open, onOpenChange }: AgentWizardProps) {
 
   const handleCopy = async () => {
     try {
-      await navigator.clipboard.writeText(installCommand);
+      // 优先使用现代 Clipboard API（需要 HTTPS 或 localhost）
+      if (navigator.clipboard && window.isSecureContext) {
+        await navigator.clipboard.writeText(installCommand);
+      } else {
+        // 降级方案：用隐藏 textarea + execCommand('copy')，兼容 HTTP 环境
+        const textarea = document.createElement('textarea');
+        textarea.value = installCommand;
+        textarea.style.position = 'fixed';
+        textarea.style.opacity = '0';
+        textarea.style.pointerEvents = 'none';
+        document.body.appendChild(textarea);
+        textarea.select();
+        const ok = document.execCommand('copy');
+        document.body.removeChild(textarea);
+        if (!ok) throw new Error('execCommand failed');
+      }
       setCopied(true);
       toast.success('安装命令已复制');
       setTimeout(() => setCopied(false), 2000);
     } catch {
-      toast.error('复制失败，请手动复制');
+      toast.error('复制失败，请手动选中复制');
     }
   };
 

frontend/src/components/server/ServerTableRow.tsx

@@ -12,7 +12,7 @@ import {
   DropdownMenuTrigger,
 } from '@/components/ui/dropdown-menu';
 import { TableCell, TableRow } from '@/components/ui/table';
-import { Edit, MoreHorizontal, Server as ServerIcon, Terminal, Trash2 } from 'lucide-react';
+import { Edit, MoreHorizontal, Server as ServerIcon, Terminal, Trash2, Download } from 'lucide-react';
 import { getStatusBadge, getTags, getUsageColor } from './serverHelpers';
 
 interface ServerTableRowProps {

frontend/src/services/api.ts

@@ -143,6 +143,10 @@ export const serverApi = {
     // 后端暂无单机 metrics，用 dashboard metrics
     return http.get('/dashboard/metrics');
   },
+  // 远程安装 Agent（通过 SSH 执行安装脚本）
+  installAgent: async (id: number | string, token: string) => {
+    return http.post<{ installed: boolean; arch: string; output: string }>(`/servers/${id}/install-agent?token=${encodeURIComponent(token)}`);
+  },
 };
 
 // ==================== K8s 集群 API ====================